VERIFIKASI METODE IN SILICO UNTUK PREDIKSI TOKSISITAS SENYAWA BAHAN TAMBAHAN PANGAN

• Main Author: Fadhil Muhammad, Ghazy
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/61920
• Institution: Institut Teknologi Bandung

Due to development of computation technology, toxicity prediction of a chemical compound can be performed using computational or in silico methods. The in silico approach to screen hazardous/toxicity properties has begun to be widely used by researchers and also be used in establishing regulations in several European countries. One of the software that can be used for toxicity prediction is VEGA. The aim of this study was to verify in silico method that can be used as initial safety assessment of the compounds proposed as food additives . The software verification was conducted using positive control and negative control compounds and then followed by assessment of test compounds. 150 food additive compounds were selected as the test compounds, which belong to the functional classes of flavours, antioxidants, sweeteners, preservatives, and colors. Verification parameters were sensitivity, specificity, accuracy, positive predictivity, negative predictivity, false positive rate, false negative rate, and receiver operating characteristic. The prediction methods that meet the acceptance requirements were then used to predict the toxicity of food additives and the results of them were compared with their experimental data. The prediction method showing the least prediction error was then considered as the best method for toxicity prediction. In hepatotoxicity, the prediction method did not meet the acceptance requirements, while for carcinogenicity, only 2 prediction methods passed the acceptance requirements, which are CAESAR and IRFMN/Antares. After comparing the two methods, IRFMN/Antares gave better results than CAESAR. In developmental toxicity, the PG method was better than the CAESAR model. In the case of skin sensitization end-point, no test compound was tested due no method meets the acceptance requirements. In terms of mutagenicity, the SarPy/IRFMN method was the best method, followed by the CAESAR , the ISS, and KNN methods.